BROOKHAVEN SCIENCE ASSOCIATES Abstract Vibrating Wire R&D for Magnet Alignment Animesh Jain, NSLS-II Project The alignment tolerance for a string of magnets on a girder in NSLS-II is ±30 microns. Such a level of alignment is difficult to meet with manufacturing tolerances and survey alone. It was decided early in the project to use the vibrating wire technique to align magnets on a girder for NSLS-II. This technique was developed at Cornell, and good measurement resolution of ~few microns was already demonstrated in quadrupoles. However, there were no systematic studies available to demonstrate the absolute accuracy of the measured centers. Similarly, very little work was done in applying this technique to sextupoles. In view of this, an R&D program was initiated at NSLS- II in early 2007. Extensive work was carried out to identify various sources of errors and means were devised to minimize such errors. A brief description of various aspects studied is presented in this talk. This R&D work culminated in building a state-of-the-art vibrating wire measurement system for NSLS-II and demonstration of absolute accuracy of ~±5 microns for sextupoles. The system is now fully operational for production measurements and over 1/3 of all the multipole girders have already been aligned to well under the required tolerance using this system. *Work performed under auspices of the United States Department of Energy, under contract DE-AC02-98CH10886 1
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
BROOKHAVEN SCIENCE ASSOCIATES
Abstract
Vibrating Wire R&D for Magnet Alignment
Animesh Jain, NSLS-II Project The alignment tolerance for a string of magnets on a girder in NSLS-II is ±30 microns. Such a level of alignment is
difficult to meet with manufacturing tolerances and survey alone. It was decided early in the project to use the vibrating wire technique to align magnets on a girder for NSLS-II. This technique was developed at Cornell, and good measurement resolution of ~few microns was already demonstrated in quadrupoles. However, there were no systematic studies available to demonstrate the absolute accuracy of the measured centers. Similarly, very little work was done in applying this technique to sextupoles. In view of this, an R&D program was initiated at NSLS-II in early 2007. Extensive work was carried out to identify various sources of errors and means were devised to minimize such errors. A brief description of various aspects studied is presented in this talk. This R&D work culminated in building a state-of-the-art vibrating wire measurement system for NSLS-II and demonstration of absolute accuracy of ~±5 microns for sextupoles. The system is now fully operational for production measurements and over 1/3 of all the multipole girders have already been aligned to well under the required tolerance using this system.
*Work performed under auspices of the United States Department of Energy, under contract DE-AC02-98CH10886
1
BROOKHAVEN SCIENCE ASSOCIATES
Vibrating Wire R&D for Magnet Alignment
Magnet WorkshopApril 11-12, 2012
Animesh Jain for the NSLS-II magnet team
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 20123
Introduction• For optimum performance, the magnetic axes of quadrupoles
and sextupoles in NSLS-II should be aligned to better than ±30 microns.
• It is difficult to achieve the required accuracy using magnet fiducialization, coupled with optical survey.
• It is difficult, and expensive, to maintain the required machining and assembly tolerances in a long support structure (~5 m) holding several magnets.
• It is desirable to achieve the required alignment using direct magnetic measurements in a string of magnets.
• The vibrating wire technique, developed at Cornell, was deemed to be the most appropriate for this task.
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 20124
Magnet Alignment R&D• Although the vibrating wire technique had been used in the past
for quadrupole measurements, very little work had been done in sextupoles. Also, little was known about absolute accuracy.
• An R&D program was initiated to further develop the technique at BNL and demonstrate the required accuracy for both quadrupoles and sextupoles.
• Good measurement reproducibility has been achieved as a result of several improvements made over the course of this R&D program, which started in January, 2007.
• The technique has now been used successfully to precision align 30 girders for NSLS-II storage ring.
• A sustained throughput of 2 girders per week is achieved.
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 20125
The Vibrating Wire Technique: Basics
• An AC current is passed through a wire stretched axially in the magnet.• Any transverse field at the wire location exerts a periodic force on the wire, thus
exciting vibrations.• The vibrations are enhanced if the driving frequency is close to one of the
resonant frequencies, giving high sensitivity. • The vibration amplitudes are studied as a function of wire offset to determine the
transverse field profile, from which the magnetic axis can be derived.
X-YStage
X-YStage
X Y
Weight
Wire carryingsinusoidal current
MagnetMover
Magnet
WireVibrationSensors
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 20126
Main Features of the BNL Vibrating Wire System• Aerotech ATS03005 stages for wire movement
(0.1 micron resolution; 2.5 micron/25 mm accuracy).
• Wire ends are defined by stainless steel V-notches.
• Set of 7 fiducials at each end to locate the V-notches.
• A pair of X-Y wire vibration sensors at each end of the wire. Allows two independent, simultaneous measurements for data verification and redundancy.
• Computer controlled piezo stages to recenter wire sensors at each scan position to minimize non-linear effects.
• Completely rewritten acquisition and analysis software with scripting support for flexibility in experiment control.
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 20127
Wire End Support (V-notch)
Fiducials relate the wire ends to the overall girder coordinate system.
Stainless V-notch
Fiducial nests (7)
Holes to help locate the notch relative to fiducials
A V-notch with radius much smaller than the wire was chosen. The wire position is thus insensitive to the actual radius of the V-notch.
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 20128
Wire Vibration Sensors
As the wire is moved horizontally or vertically, the position of the wire relative to the sensor changes slightly (~ a few microns) due to imperfections in the stage motion. This causes a change in the operating point of the sensor.
An automated piezo stage was added to keep the wire “centered” in the sensors during a scan.
Coarse manual adjustment in orthogonal axis
X-Sensor
Fine, automated adjustment along measurement axis using piezo stages
Y-Sensor
The wire motion sensors are inexpensive photointerrupters(Model GP1S094HCZ0F)
A pair of sensors is located on both ends of the wire, thus allowing two simultaneous measurements.
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 20129
Complete Wire Mover AssemblyCamera to ensure wire is correctly seated in the notch
Light Shades to reduce noise from stray light
V-notch holder with fiducials
A similar assembly is present at the other end of the wire, except that the pulley and weight are replaced by a fixed wire end.
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 201210
Alignment Issues Studied• Wire sag correction (presented at IMMW15)• Detector sensitivity to orthogonal motion• Sensitivity to Yaw/Pitch of magnets• Accuracy of quadrupole center measurement• Accuracy of sextupole center measurement• Background field correction• Ability to precisely move magnets and secure to the girder.• Reproducibility of the girder vertical profile.• Stability of magnet alignment during transportation and
handling of the girder.
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 201211
Detector Sensitivity to Orthogonal Motion
Measured signal is contaminated by sensitivity to motion along the orthogonal axis, thus causing errors in the measurement of magnetic center.
A rigorous analysis has shown that the error in horizontal/vertical center is minimized if wire is scanned at the vertical/horizontal center in both quadrupoles and sextupoles. This implies that a rough center must be found first before the final scan.
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 201212
Resonant Mode to Use: Yaw/Pitch Sensitivity
• Should have a maxima near the axial center of the magnet being measured.
• Preferably even numbered modes should be used to avoid contribution from any axially uniform background fields (e.g. earth’s field).
• It may be impractical to find a mode with maxima exactly at the axial center for every magnet on the girder. This causes sensitivity to yaw and pitch.
• One should choose a mode that minimizes sensitivity to yaw and pitch without sacrificing signal strength.
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 201213
Sensitivity to Yaw and Pitch
Signal for mode = n is proportional to Byn(for a given detector position)
In a quadrupole magnet with offset x0 and yaw angle , located at z = zmag:
Error in center determination due to yaw:
A similar analysis for sextupoles is much more tedious, but the same expression for x0 is obtained in the end!
A similar expression applies to vertical offset and pitch angle.
zmag = magnet axial position; G = Gradient
Lmag = magnetic length
L = wire length
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 201214
Yaw/Pitch Sensitivity in Quad MeasurementsYaw/Pitch Sensitivity in Quad Measurements
-40
-30
-20
-10
0
10
20
30
40
0 2 4 6 8 10 12 14 16Mode Number
Yaw
/Pitc
h Se
nsiti
vity
(m
icro
n/m
rad)
m 23.0m 98.3
m 27.7
mag
magLzL
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 201215
Sextupole Measurements Using By and Bx
• Obtaining centers from By vs. x and By vs. y plots uses only one set of sensors, and requires quadratic fits.
• One could also use scans of Bx vs. x (or y) for various values of y (or x). These plots are expected to be linear with slopes proportional to offsets in y (or x) direction.
• Doing three such scans allows to obtain centers from both Bx and By data. With 2 sets of sensors, one gets four values of magnetic center.
2
20
20
3)()(
refy
RyyxxBB
2
003
))((2ref
xR
yyxxBB
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 201216
Yet Another Way to Measure Sextupoles
Fit measured data to a truncated Fourier Series, giving quadrupole and sextupole terms.
Both X and Y Centers can be obtained using any of the 4 detectors in a single scan.
Circular Scan (R = 1 mm) in SLS Sextupole at 100A
-60
-40
-20
0
20
40
60
0 45 90 135 180 225 270 315 360Angle (deg.)
Sign
al (a
rbitr
ary
units
)
X1 X1Fit Y1 Y1FitX2 X2Fit Y2 Y2Fit
19-Mar-2008Uncorrected
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 201217
Comparison of Sextupole Data Using Bx and By
Summary of Sextupole Center Measurements in SR110 at 100A19-Mar-2008: Comparison of Circular and Linear Scan Results
Uncorrected Background Corrected Correction Needed inMethodNumber Scan Type Sensor
Used
CIRCULAR(single scan;
25 points)
LINEAR(6 scans;7 points
each)
Mean of all 8 methods:
Std. Deviation of all 8 methods:
Circular to Linear Scan Diff.:
Mean of Circular Scan Data:
Std. Deviation of Circular Scan:
Mean of Linear Scans:
Std. Deviation of Linear Scans:
Consistency comparable to quadrupole measurements is obtained in more recent data.
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 201218
Issue of Background Fields in Sextupole Meas.
• There is a significant quadrupole background field from quadrupole magnet(s) even when these are unpowered.
• Based on rotating coil data, the remnant integrated quadrupole field could amount to a change in horizontal center by hundreds of microns, depending on quad position and the mode used for sextupole measurements.
• The vertical center measurement is not affected because By
(or Bx) is independent of y (or x) in a normal quadrupole field.
• Effectiveness of background correction has been tested by measuring a sextupole in the presence of a quadrupole which was either unpowered, or was powered at 2 A (apparent center shift of ~600 microns). Corrected center after background subtraction was within ±5 microns of the value without large background.
BROOKHAVEN SCIENCE ASSOCIATES
NSLS-II Magnet Workshop: April 11-12, 201219
Alignment StabilityC26G4 returned from tunnel (Move 4_Repeated2)